Decision Transformer (DT) is a Transformer-based architecture originally designed for Reinforcement Learning (RL), in which decision-making is formulated as a sequence modeling problem through return-to-go conditioning. DT has been applied in different fields, in which intelligent agents have to undergo an intensive offline process. Yet, its applicability to supervised classification tasks has barely been studied. This study evaluates the performance of DT in a classification problem without explicit temporal dependencies, using the Iris Flower Dataset as a case study. To this end, its accuracy was compared with that of a conventional classifier, using Logistic Regression (LR) as a reference. Multiple experiments were conducted on different versions of the dataset to analyze the model’s consistency. The results show that DT outperforms LR in terms of accuracy in the evaluated scenarios, validating its feasibility as a classifier.